-
The importance of ant-rich habitats for thepersistence of the
Wryneck Jynx torquilla on farmland
MURIELLE MERMOD,1 THOMAS S. REICHLIN,1,2 RAPHAËL ARLETTAZ1,3
& MICHAEL SCHAUB1,2*
1Institute of Ecology and Evolution – Division of Conservation
Biology, Baltzerstrasse 6, CH-3012 Bern, Switzerland2Swiss
Ornithological Institute, CH-6204 Sempach, Switzerland
3Swiss Ornithological Institute, Valais Field Station, Nature
Centre, CH-3970 Salgesch, Switzerland
The frequency of territory occupancy is a good indicator of
territory quality. We studiedterritory occupancy in a Swiss
population of the Wryneck Jynx torquilla, a decliningfarmland
woodpecker, with the aim of identifying key habitat features for
conservationmanagement. Both static and dynamic approaches were
applied using data on nest-siteoccupancy of 100 territories from
six successive years. The static approach models theprobability of
territory occupancy; the dynamic approach estimates territory
colonizationand extinction. Frequently occupied territories were
settled earlier in the season, suggest-ing that they may be of
better quality, and birds settling in these territories had
higherbreeding success. Probability of territory occupancy
increased with the area of old pearorchards and decreased with the
area of vegetable cultivation. Both the area of old pearorchards
and the presence of conspecifics within a territory were positively
related toterritory colonization, whereas territory extinction was
negatively related to habitatheterogeneity. Old pear orchards were
characterized by having both the highest densityof ant nests and
the greatest amount of bare ground. The latter is likely to
facilitateaccess to ant prey. To ensure persistence of Wryneck
populations in farmland, hetero-geneous habitat matrices with high
ant nest density and bare ground should bepromoted. Finally,
provision of artificial nesting cavities is likely to enhance
territoryoccupancy. Providing that these key resources are present,
Wrynecks are likely to persisteven in intensively farmed areas.
Keywords: ants, colonization and extinction probabilities, food
availability, habitat selection, occu-pancy model, territory
quality.
Individuals preferentially select breeding territoriesin
high-quality habitat patches that provide suit-able resources such
as food, breeding sites andshelter from predators because
reproductive out-put usually increases with increasing territory
qual-ity (Andrén 1990, Tye 1992, Holmes et al. 1996,Pärt 2001).
Territory selection is therefore crucialto the reproductive fitness
of individuals. Accord-ing to the ideal despotic distribution, the
highestquality territories are selected first (Fretwell &Lucas
1969). The occupancy of territories is thus anon-random process,
with the best quality territo-ries being monopolized by the
strongest individuals
or the first to arrive. If individuals are distributedaccording
to the ideal despotic distribution, thefrequency of territory
occupancy will be positivelycorrelated with territory quality
(Krüger 2002,Sergio & Newton 2003, Sim et al. 2007),
withlow-quality territories only occupied when breed-ing density is
high. Thus, territory variables thatare related to the frequency of
territory occupancymay be good measures of habitat quality.
This static view of territory occupancy can beextended to a
dynamic occupancy approach.Territory occupancy involves two
processes:colonization and extinction. These local processesmay
depend on stochastic, intrinsic or extrinsicfactors (Hanski 1998),
such as habitat quality orconspecific attraction (Stamps 1988,
Muller et al.
*Corresponding author.Email: [email protected]
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
Ibis (2009), 151, 731–742
-
1997, Danchin et al. 1998). Territory variables thatare
positively correlated with colonization andnegatively with
extinction indicate high quality.
A key issue in conservation ecology is the identi-fication of
territory quality gradients, to strategi-cally prioritize
conservation management. The goalof this study was to identify
those habitat featuresthat are the best indicators of habitat
quality forWrynecks Jynx torquilla. This species is a second-ary
cavity breeder that feeds almost exclusively onground-dwelling ants
(Freitag 1998). Wrynecksoccur in different kinds of semi-open
habitats, suchas farmland or open woods, but their populationsare
declining throughout Europe (Sanderson et al.2006).
We first mapped a number of habitat features inrecorded and
potential Wryneck breeding territo-ries. We then assessed the
abundance of ant nestsin territories, as ant broods constitute the
staplefood provisioned to Wryneck nestlings (Bitz &Rohe 1993,
Freitag 1998). Finally, we related habi-tat type, food availability
and the presence of con-specifics to the observed pattern of
territoryoccupancy. Analyses were carried out using bothstatic
(variables related to frequency of occupancy)and dynamic occupancy
models (variables relatedto local colonization and extinction
processes of agiven territory) to identify associations that
couldlead to the development of conservation manage-ment
recommendations for farmland Wryneckpopulations.
METHODS
Study site and study species
The study area was located in the plain of theRhône valley,
between Vernayaz and Sion (southwest Switzerland, 46�13¢N, 7�22¢E;
482 m asl).The area is characterized by intensive
farming,consisting mainly of plantations of dwarf fruit
trees(hereafter orchards) and vegetable cultivation. Thestudy area
covered 62 km2, within which 351 nest-boxes were installed in 2002
at 195 locations(mostly inside agricultural shacks, most
buildingshaving two boxes each). These nestboxes werespread
regularly in apparently suitable semi-openhabitats across the study
area and their numberremained constant during the study period.
Natu-ral cavities as well as nestboxes were also presentbut
appeared to be scarce, although their numbersare unknown.
In contrast to most other woodpeckers, Wrynecksare
secondary-cavity breeders. They also require for-aging grounds
offering sparse vegetation cover, facili-tating access to ant nests
(Hölzinger 1992, Bitz &Rohe 1993, Freitag 1996, N. Weisshaupt,
R. Arlettaz,T.S. Reichlin, A. Tagmann-loset, M. Schaub,
unpubl.data). In our study area, 90% of the food provi-sioned to
Wryneck nestlings comprised ant larvaeand nymphs (Freitag 1998).
Telemetry studies inthe study area revealed that Wrynecks
foragemostly within 100–125 m of their nest-site, havingmedian
home-ranges of 3.9 ha (N. Weisshaupt,R. Arlettaz, T.S. Reichlin, A.
Tagmann-loset,M. Schaub, unpubl. data). Orchards and fallow landare
the preferred foraging habitats, and preferredfeeding locations
have more than 50% cover of bareground (N. Weisshaupt, R. Arlettaz,
T.S. Reichlin,A. Tagmann-loset, M. Schaub, unpubl. data).
Between 2002 and 2007, all 195 nest-sites werechecked
fortnightly during the breeding season.Once detected, a brood was
monitored every 3–4days. A brood was defined as any clutch
containingat least one egg, irrespective of the outcome, whilsta
successful brood yielded at least one fledgling.
Design and habitat variables
A random sample of 100 nest-sites was selectedfrom the 195
available sites in the study area.Around each nest-site we
delineated a circle of 111-m radius, to define a 3.9-ha area
equivalent to themedian foraging home-range size (N. Weisshaupt,R.
Arlettaz, T.S. Reichlin, A. Tagmann-loset,M. Schaub, unpubl. data).
In only three cases wasthere an overlap between two adjacent
estimatedhome-ranges (maximal overlap of 8.2%). Of the100
nest-sites, 62 were occupied in at least 1 year.The habitat
characteristics of the selected territorieswere mapped in early
summer 2007 (SupportingInformation Fig. S1). Variables recorded
werehabitat type, number of trees and percentage coverof bare
ground, these being the features identi-fied as important by N.
Weisshaupt, R. Arlettaz,T.S. Reichlin, A. Tagmann-loset, M. Schaub
(unpubl.data). The percentage of bare ground was estimatedvisually
for each cropping unit (parcel) in the field.
Occupancy models assume that territory qualityand food resources
remain constant over time andthat individuals are able to locate
the best qualityterritories promptly. In our study area,
territoryquality was assumed to be constant over the 6-yearperiod,
as orchards, which covered on average
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
732 M. Mermod et al.
-
48.9 ± 0.6% of foraging territories (n = 100), havea slow
replacement turnover, and because ant nestsare relatively long use
(Seifert 1996). In addition,there was an excess of available
nest-sites (on aver-age 2.03 ± 0.06 nestboxes per territory), so
weassumed that territory selection would operateindependently of
nest-site availability.
The density of ant nests within a territory wasalso used as a
potential criterion for estimatinghabitat quality. Perches such as
trees, poles, shacksand fences are an important habitat feature
forWrynecks, as perches enhance detectability of antnests. We thus
distinguished between ant nestavailability with perches (orchards,
vineyards, river-bank and pasture) and without perches
(vegetablecultivation, fallow land and meadow). Nests fromwoodland
ant species (Formica spp. and Campono-tus spp.) were not included
in the food availabilityestimates because these species are almost
nevereaten by Wrynecks (Seifert 1996, Freitag 1998).Ant nest
accessibility was accounted for in themodelling, as this depends
primarily on the area ofbare ground in the vicinity of an ant nest
andbecause Wrynecks capture and eat prey on theground (N.
Weisshaupt, R. Arlettaz, T.S. Reichlin,A. Tagmann-loset, M. Schaub,
unpubl. data). Thequadratic terms of the three discrete
variablesquantifying bare ground with perches, bare groundwithout
perches and total bare ground were alsoused to identify optimal
proportions of these keyhabitat components. Finally, an index of
potentialcompetition with Hoopoes Upupa epops, whichuse the same
nestboxes as Wrynecks, was derivedfrom the number of years between
2002 and 2007that a Hoopoe pair had occupied one of the
twoavailable nestboxes within a territory.
In the dynamic occupancy analysis, we alsoincluded a binary
variable indicating presence ofconspecifics in a given year (scored
1 if there wereother breeding Wrynecks within a 500-m radius,twice
the maximal foraging distance, of the focalnest-site). Conspecifics
may influence site occu-pancy through patterns of social
attraction(Danchin et al. 1998).
Spatial data were digitized using ARCGIS 9.1(ESRI, Redlands, CA,
USA). In total, 2589 differ-ent cropping units were recorded and
their areascalculated from digitized polygons. Multiple unitsof the
same habitat type within a territory weresummed and their
proportional area was usedin the analysis. Total percentage of bare
groundwithin a territory was calculated by summing the
percentage of bare ground per unit, weighted byits proportion of
the entire territory area.
Food availability: ant nest abundanceand accessibility
To estimate overall food availability within a terri-tory,
habitat-specific ant nest densities were multi-plied by their
proportional area of territory andsummed. Density of terrestrial
ant nests wasassessed throughout the study area in 2003 and2004.
Sampling was based on a stratified design (atleast 90 randomly
selected plots per main habitattype: orchards, vineyards, meadow,
riverbank, fal-low land and vegetable cultures). From 2005 to2007,
assessments were restricted to orchards (withdifferent fruit
types), as these were identified as themost important foraging
habitat (N. Weisshaupt,R. Arlettaz, T.S. Reichlin, A.
Tagmann-loset,M. Schaub, unpubl. data). Sample size variedbetween
habitat types and years, ranging from 10to 175. Each sampling
location was situated in thecore of a given crop type to avoid edge
effects. Antnests were surveyed in five 2-m2, randomly
definedreplicates, by scraping the topsoil with a rake andcounting
the nests. Surveys took roughly 5 min perreplicate, and were always
carried out in the firsthalf of May on 3–9 consecutive days, under
similarweather conditions. Habitat type (for orchards, alsoage and
fruit type), percentage of bare ground, veg-etation height, and
number and relative size of antnests were recorded. An ant nest was
defined bythe presence of an aggregation of ‡ 20 imagos, orthe
presence of eggs or larvae. A few individualsfrom each nest were
sampled for subsequent spe-cies identification (Della Santa 1994,
Seifert 1996).
Ant nests that were small or located deep in thesoil could
easily be overlooked. Therefore, we esti-mated detection
probability of ant nests in an addi-tional study in 2007, and
corrected the ant nestcounts accordingly (Supporting
InformationAppendix S1). We repeatedly searched for antnests at
exactly the same five 2-m2 plots in 19 ran-domly selected orchards
(95 replicates), using thesame technique as for the ant nest
density assess-ment described above. All 95 replicates wererecorded
over a period of 3–4 days, during threerecording sessions in early
May, June and July. Datafor mean daytime temperature, measured 5
cmabove ground, were obtained from MeteoSwiss(on-line database). A
few ants were sampled fromeach nest for subsequent species
identification.
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
Territory quality in Wrynecks 733
-
The resulting detection histories for each ofthe 95 replicates
were analysed with MARK(White & Burnham 1999), using occupancy
models(MacKenzie et al. 2002). We did not distinguishbetween
detection probabilities of different antspecies, as sample sizes of
different species wereunbalanced. Vegetation height (range: 0–110
cm),percentage of bare ground (range: 0–100%) andmean hourly
temperature (range: 3.3–35.0 �C)were used to model detection
probability. Allvariable combinations were tested with both
aconstant and a time-dependent intercept, giving atotal of 16
models. The occupancy probability wasalways kept constant. Models
were ranked accord-ing to Akaike’s Information Criterion adjusted
forsmall samples (AICc) and the corresponding AICcweights (Burnham
& Anderson 1998). Model aver-aging was performed for the
smallest subset ofmodels with accumulated AICc weights summingto
0.95, to account for model selection uncertainty(Burnham &
Anderson 1998).
The accessibility of ant nests for Wrynecks isprimarily
determined by the amount of bareground (N. Weisshaupt, R. Arlettaz,
T.S. Reichlin,A. Tagmann-loset, M. Schaub, unpubl. data). Asan
indication of food accessibility, we thereforeused average amount
of bare ground for each habi-tat, estimated with a linear mixed
effects model,with territory as a random effect. Estimates
andconfidence intervals were obtained by bootstrap-ping, with 1000
repetitions.
Occupancy analyses
Assumptions
According to the ideal despotic distributionhypothesis (Fretwell
& Lucas 1969), high-qualityterritories are settled earlier, and
thus more fre-quently occupied territories should be
settledearlier. We used laying date of the first egg as anindex of
territory settlement. To test whethersettlement order depended on
the year, we usedlinear mixed models and a likelihood-ratio test.
Wethen related the median of the laying date of thefirst egg for
each territory to the frequency of itsoccupancy using linear
regression. Only firstbroods were considered (n = 108).
If breeding success is site-dependent, it shouldbe positively
correlated with frequency of territoryoccupancy. This prediction
was tested using linearmixed models with Poisson and binomial error
dis-tributions, fitting three components of breeding
success of a territory (clutch size, number of fledg-lings,
probability of successfully raising a brood) asdependent variables,
territory as a random effect,and frequency of territory as
occupancy and yearas fixed independent variables. Sample sizes
differedbetween analyses because only complete clutches(defined as
a least one egg hatched) for the clutchsize analysis and only
successful broods for the anal-ysis of the number of fledglings
were considered(clutch size: n = 122 broods from 53
territories;number of fledglings: n = 94 broods from 50
terri-tories; proportion successful: n = 175 broods from62
territories).
Static model
To model the annual probability that a territorywas occupied, we
used a logistic regression modelwith a binomial error distribution.
The numeratorof the response variable was the number of times
aterritory was occupied, and the denominator wasthe number of study
years (n = 6). From allrecorded territory variables (Supporting
Informa-tion Tables S1 and S2), some were excluded fromthis
analysis either due to their irrelevance as for-aging habitat
(anthropogenic habitat and water), orbecause they occurred in fewer
than 20 territories(riverbank and pasture). The remaining 22
vari-ables were tested for pair-wise correlation, usingSpearman’s
rank correlation test. As a result, thevariable relating to ant
nests in habitats withoutperches was excluded, as it was highly
correlated(rs > |0.7|) with two other variables. The
quadraticterms of the variables relating to total bare ground,bare
ground with perches and bare ground withoutperches were also
included, to test for the presenceof an optimum, giving 24
explanatory variables.We then fitted univariate models containing
eachof these explanatory variables and ranked themaccording to
their AICc values. All variables frommodels with a DAIC < 4 when
compared with thebest model were included into the second
model-ling step. We built models with all possible combi-nations of
the remaining variables (four variables;15 models) and ranked them
according to theirAICc values. Predictions were made using
modelaveraging, including all models within the secondstep of the
analysis that summed to at least 0.95 ofthe AICc weight.
Dynamic model
To model colonization and extinction probabilities,we used a
dynamic occupancy model (MacKenzie
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
734 M. Mermod et al.
-
et al. 2003, Royle & Dorazio 2008) fitted with theprogram
MARK (White & Burnham 1999). Thismodel is parameterized with
initial occupancy, col-onization and extinction probabilities. The
same24 variables were used as for the static occupancymodel. In
addition, we considered the binary vari-able relating to the
presence of conspecifics. Modelselection was performed in multiple
steps: first,univariate models were fitted containing each ofthese
25 variables for extinction and colonizationprobabilities, with a
constant and a time-depen-dent intercept. All models with a DAIC
< 4 wereselected for inclusion in the extinction and
coloni-zation modelling. In the second step, the variablesfrom the
selected extinction models werecombined with the variables from the
selectedcolonization models. From the selected variables,at most
one variable for extinction and one forcolonization were combined,
using both time-dependent and constant intercepts (36 models).Model
averaging was performed for modelsaccounting for 0.95 of the AICc
weight. The initialoccupancy probability was not modelled
againstexplanatory variables.
Breeding success
To assess whether the important territory variablesidentified in
the static and dynamic approacheswere linked to breeding success,
we fitted equallystructured generalized linear mixed models asused
for the relationship between breeding successand frequency of
occupancy. The dependent vari-ables were clutch size, number of
fledglings fromsuccessful broods and proportion of
successfulbroods. Territory identity was entered as a
randomvariable and year as a fixed effect. All possiblevariable
combinations using different territoryvariables were tested, with
models ranked basedon DAICc.
All statistical analyses were performed using theprogram R
version 2.5 (R Development Core Team2004), unless stated
otherwise.
RESULTS
Food supply: ant nest abundance andaccessibility
During the three visits to orchards carried out in2007, 242 ant
nests were located belonging to fourspecies: Lasius niger (75.6%)
was by far the most
frequently recorded species; L. flavus (14.9%),Tetramorium
caespitum (7.4%) and Solenopsis fugax(2.1%) were less common.
(a)
(b)
(c)
Figure 1. Relationship between detection probability of ant
nests and temperature (a), vegetation height (b) and amount
of
bare ground (c). Shown are model-averaged predictions
(based on results in Supporting Information Table S3) with
95% confidence intervals.
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
Territory quality in Wrynecks 735
-
Detection probability increased with tempera-ture, decreased
with vegetation height, increasedslightly with amount of bare
ground (Fig. 1) anddeclined with season (mean ± se; May: 0.723
±0.058; June: 0.588 ± 0.061; July: 0.438 ± 0.059)(see Supporting
Information Table S3).
To estimate the detection probability of antnests for each
replicate, we used the model aver-aged parameter estimates from May
and thereplicate-specific variables. Ant nest densitiescorrected
for detection probability were lowestin vegetable cultures and
highest in orchards,especially in pear orchards (Fig. 2).
The amount of bare ground differed signifi-cantly between
habitat types (Fig. 2). Bareground cover was moderate in orchards,
withpear orchards having more than other orchards.The combination
of both high ant density and alarge amount of bare ground (i.e.
high accessibil-ity) was highest in orchards, vineyards and fal-low
land.
Occupancy analyses
Assumptions
The most frequently occupied nest-sites were set-tled earlier in
the season (estimate = )3.298 ±0.894, P < 0.001), and settlement
order was
independent of year (likelihood-ratio test,v21 = 1.726, P =
0.189). Clutch size (estimate =0.011 ± 0.127, P = 0.921) and number
of fledg-lings (estimate = 0.011 ± 0.149, P = 0.907) werenot
associated with the frequency of territoryoccupancy, but the
probability that a brood wassuccessful increased significantly with
the fre-quency of occupancy (estimate = 0.057 ± 0.029,P =
0.047).
Static model
In univariate models, cover of old pear orchardsshowed the
strongest impact on occupancy (Sup-porting Information Table S4).
The next best vari-ables, bare ground with perches squared,
vegetablecultures and ant nests from area with perches,were still
within 4 DAICc units, so they wereincluded in the second step.
Ant
nes
t den
sity
(N
/m2 )
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Veg
etab
les
Vin
eyar
d
Mea
dow
Fal
low
land
Riv
erba
nk
App
le
Apr
icot
Pea
r0
10
20
30
40
50
60
Am
ount
of b
are
grou
nd (
%)
Ant densityAmount of bare ground
110
184
90 115
270
147
210
105
150
37 520
524
155
127
405
255
Figure 2. Estimated ant nest densities and amount of bare
ground per habitat type obtained by fitting linear mixed
models.
Shown are the estimates for each habitat type with 95%
confi-
dence intervals. Ant densities are corrected for imperfect
detection probability. The number of samples for each
habitat
type is given at the bottom of each bar.
Table 1. Summary results of the static occupancy modelling
of
Wryneck territories, when the three explanatory variables
selected in the first step (Supporting Information Table S4)
are
used together. Given are the DAICc, AICc weights (wi),number of
parameters (K) and residual deviance. The models
are ranked according to their AICc weight.
Model DAICc wi K Deviance
Old pear orchard + vegetable
cultures
0.00 0.304 3 218.52
Old pear orchard + vegetable
cultures + bare ground with
perches2
1.61 0.136 5 216.12
Old pear orchard + bare ground
with perches21.64 0.134 4 218.16
Old pear orchard + vegetable
cultures + ant with perches
1.99 0.112 4 218.51
Old pear orchard + vegetable
cultures + ant with perches
+ bare ground with perches2
2.55 0.085 6 215.06
Old pear orchard 3.17 0.062 2 223.69
Old pear orchard + ant with
perches + bare ground with
perches2
3.40 0.055 5 217.92
Old pear orchard + ant with
perches
3.42 0.055 3 221.93
Vegetable cultures + bare
ground with perches26.28 0.013 4 222.79
Bare ground with perches2 6.48 0.012 3 225.00
Vegetable cultures 6.81 0.010 2 227.33
Ant with perches 6.91 0.010 2 227.43
Ant with perches + bare ground
with perches28.04 0.005 4 224.56
Vegetable cultures + ant
with perches + bare ground
with perches2
8.23 0.005 5 222.75
Constant model 12.43 0.001 1 234.94
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
736 M. Mermod et al.
-
The second step showed that the combinationof old pear orchards
and vegetable culturesresulted in the most parsimonious model(Table
1). Model-averaged occupancy probabilitiesincreased with the
proportion of old pear orchardsand decreased with the proportion of
vegetables(Fig. 3). Occupancy was highest when the coverof bare
ground was around 30%, but remainedalmost constant with increasing
number of antnests.
Dynamic model
The most parsimonious set of univariate models(DAICc < 4)
affecting extinction probabilityincluded number of cropping units
and cover ofyoung apple orchards (Supporting InformationTable S5).
The most parsimonious set of modelsaffecting colonization
probability included cover ofold pear orchards and presence of
conspecifics(Supporting Information Table S6). In addition,
colonization probability differed across study years,whereas
extinction probability did not.
Multivariate models of extinction probabilityretained only
number of cropping units, whereasthe most parsimonious multivariate
model ofcolonization retained only cover of old pearorchards (Table
2). In addition, there was somesupport for an effect of the
presence of conspecif-ics on colonization probability but almost
none forthe cover of young apple orchards on
extinctionprobability.
Model-averaged probabilities of territory extinc-tion decreased
with increasing number of croppingunits and increased slightly with
the proportion ofyoung apple orchards within a territory (Fig.
4a,b).Territory colonization probability strongly increa-sed with
increasing proportion of old pearorchards within the territory
(Fig. 4c). It was alsoslightly enhanced when conspecifics were
present(Fig. 4d). There was no significant relationship
(a) (b)
(c) (d)
Figure 3. Predicted model-averaged probability of Wryneck
territory occupancy, calculated from the best models accounting for
0.95
of the AICc weight (Table 1), in relation to the proportion of
old pear orchards (a), vegetable cultures (b), ant nests from area
with
perches (c) and amount of bare ground with perches (d). The
figures show averaged estimates with 95% confidence intervals.
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
Territory quality in Wrynecks 737
-
between any measure of breeding success and thecover of old
orchards or the number of croppingunits (Table 3).
DISCUSSION
The most frequently occupied Wryneck territorieswere settled
earlier in the season and theprobability that a territory was
occupied or newlycolonized was positively related to the
proportionof old pear orchards within the territory. Theprobability
that a territory was abandoned wasnegatively related to the number
of cropping unitswithin the territory (i.e. a structure mosaic
effect).Pear orchards had the highest density of ant nests,thus
offered the highest density of food resources,and were sparsely
vegetated, which enhanced antdetection ⁄ accessibility. Finally,
territories whereconspecifics were present in the vicinity
wereslightly more likely to be colonized. Breedingsuccess, however,
was not related to any of thesevariables.
Territory selection and territory quality
Territory occupancy can be regarded as the resultof two
selection processes. The first selection sepa-rates locations in
which essential resources arelacking from locations in which all
essentialresources are available. The second selectionfollows the
gradient of habitat quality. In ouranalysis we did not
differentiate between thesetwo processes because we regarded all
studiedareas as potentially suitable for Wrynecks, assum-ing that
no essential resource was lacking. Nest-boxes were installed only
in areas that wereconsidered to be suitable, semi-open habitat.
Fur-thermore, nesting cavities are a key resource forWrynecks
(Coudrain 2009), and all our territoriesharboured at least two
nesting cavities.
The proportion of old pear orchards within aterritory appears
here to be an important factoraffecting territory colonization and
occupancy. In aprevious study of the same population, old
fruitorchards, but not necessarily old pear orchards,had been
determined to be preferred foraging habi-tats of Wrynecks (Freitag
1998, N. Weisshaupt,R. Arlettaz, T.S. Reichlin, A. Tagmann-loset,M.
Schaub, unpubl. data). In general, fruit treeorchards had higher
densities of ant nests thanother habitat types, with pear orchards
harbouringhigher nest densities than apple or apricot orch-ards.
Pear orchards also had a higher proportion ofbare ground than apple
and apricot orchards, andold pear orchards had a slightly higher
proportionof bare ground (39%) than middle-aged (32%) andyoung pear
orchards (34%). Thus, pear orchardsare a favoured habitat type due
to high food den-sity and good prey accessibility. The
negativeimpact of the proportion of vegetable cultivationis in line
with the findings of Freitag (1998)and N. Weisshaupt, R. Arlettaz,
T.S. Reichlin,A. Tagmann-loset, M. Schaub (unpubl. data),
whoobserved an avoidance of vegetable cultivation byforaging
Wrynecks. Although areas of vegetableproduction were sparsely
vegetated, which wouldallow easy access to ant nests, they had a
low antnest density and were mostly devoid of perches.Thus, these
areas were not suitable foraging habi-tats and negatively affected
territory occupancy.
Territory extinction probability was negativelyaffected by the
number of cropping units, whichindicates a positive effect of
habitat heterogeneityin the agricultural matrix. According to
Dauber andWolters (2004), most ant species can experience
Table 2. Summary results of extinction (e) and colonization
(c)modelling for Wryneck territories. Shown are the 12 best
models from originally 36 fitted models. Given are the
DAICc,AICc weights (wi), number of parameters (K ) and residual
deviance. A constant intercept is indicated with (.) and a
time-
dependent intercept with ‘year’. The initial occupancy
probability (W) is constant for every model. See
SupportingInformation Tables S5 and S6 for the first modelling
steps.
Model DAICc wi K Deviance
e (number of parcels)c (year + old pear orchard)
0.00 0.439 9 595.39
e (number of parcels)c (year + presence of conspecifics)
2.92 0.102 9 598.31
e (year + number of parcels)c (year + old pear orchard)
2.96 0.100 13 590.03
e (.) c (year + old pear orchard) 3.70 0.069 8 601.15e (number
of parcels)c (presence of conspecifics)
3.84 0.064 5 607.43
e (young apple orchard)c (year + old pear orchard)
3.97 0.060 9 599.35
e (number of parcels)c (old pear orchard)
5.88 0.023 13 592.95
e (year number of parcels)c (year + presence of
conspecifics)
6.63 0.016 8 604.07
e (.) c (year + presenceof conspecifics)
6.69 0.016 9 602.07
e (year + number of parcels)c (presence of conspecifics)
6.89 0.014 9 602.27
e (young apple orchard)c (year + presence of conspecifics)
7.26 0.012 12 596.42
e (year) c (year + old pear orchard) 7.31 0.011 13 594.38
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
738 M. Mermod et al.
-
a positive edge effect, which would lead to differ-ences in
abundance between the centre and theedge of a culture parcel. This
may provide a func-tional, trophic explanation for why
Wrynecksselect heterogeneous habitats (Roth 1976, Bentonet al.
2003). Moreover, territories within highlystructured farmland
matrices are less likely to beaffected by all sorts of habitat
disturbanceoccurring at single cropping units (e.g. removal offruit
tree orchards, pesticide application) and maythus provide more
stable food resources. This maybe especially important in areas
with intensivelymanaged agriculture.
In contrast to our expectations based on a previ-ous study of
the same population (N. Weisshaupt,R. Arlettaz, T.S. Reichlin, A.
Tagmann-loset,M. Schaub, unpubl. data), the proportion of
bareground within a territory had only a marginaleffect on
territory occupancy, and none for
extinction or colonization, despite the fact thatWrynecks
preferentially forage in areas with morethan 50% bare ground (N.
Weisshaupt, R. Arlettaz,T.S. Reichlin, A. Tagmann-loset, M.
Schaub,unpubl. data). This discrepancy is most probablydue to a
scale effect (Orians & Wittenberger 1991,George & Zack
2001). At the micro-habitat scale,the proportion of bare ground
appears to beimportant, whereas at a larger spatial scale, itmight
not be that crucial.
The presence of conspecifics positivelyaffected colonization
probability, which is inaccordance with many dispersal studies
(e.g.Stamps 1988, Muller et al. 1997). Whereasexperienced birds
often rely on their own repro-ductive success to assess territory
quality, whicheventually leads to site fidelity (Schaub &
vonHirschheydt 2009), juveniles and unsuccessfulbreeders may rely
either on reproductive success
(a) (b)
(c) (d)
Figure 4. Predicted model-averaged probabilities of Wryneck
territory extinction and colonization from the best models
accounting for
0.95 of the AICc weight (Table 2). Shown are the number of
parcels (a) and proportion of young apple orchards within the
territory (b)
in relation to extinction probability, and the proportion of old
pear orchards within a territory (c) and the presence of
conspecifics within
a radius of 500 m to the nest-site (d) in relation to
colonization probability. For time-dependent models, the
predictions shown are for
the year 2004. The figures show averaged estimates with 95%
confidence intervals.
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
Territory quality in Wrynecks 739
-
of other individuals (Doligez et al. 1999) or onthe presence of
conspecifics to evaluate it(Muller et al. 1997).
High-quality territories should provide a fitnessbenefit for the
territory holder, most likely in termsof increased breeding success
(Andrén 1990,Holmes et al. 1996, Pärt 2001). In the presentstudy,
breeding success of Wrynecks was notrelated to habitat variables
identified as importantand only marginally to the frequency of
territoryoccupancy, pointing towards a possible mismatchbetween
habitat preference and fitness (Arlt & Pärt2007). One likely
reason for this result is that thebreeding success of Wrynecks has
a stochasticcomponent because it is affected by the
prevailingweather (Geiser et al. 2008). This might blur
anyrelationship with habitat variables. A furtherexplanation is
that the considered components ofbreeding success are not the
relevant fitness com-ponents. Other components such as
nestlinggrowth, and post-fledging and adult survival maybe more
strongly affected by territory quality.
Implications for conservation
Throughout its range, the Wryneck occurs in vari-ous
savannah-like landscapes such as light wood-land, traditional
orchards, parks and vineyards(Glutz von Blotzheim & Bauer
1980). In our studyarea, intensively farmed pear orchards seem
tooffer the best foraging conditions for Wrynecks,offering a high
density of ant nests and optimalprey detectability ⁄ accessibility.
The reduction offood availability due to increasing density
ofground vegetation cover affecting accessibility toants or due to
the reduction of ant abundance,both as a result of intensive
farming, has undoubt-edly strongly contributed to the large-scale
declineof Wryneck populations. Farming practices thatmaintain or
create sparsely vegetated patches (e.g.through stopping fertilizer
application, mechanicalor chemical destruction of ground
vegetation,removal of topsoil) within savannah-like land-scapes are
likely to enhance Wryneck populations.Although the Wryneck is often
viewed as a speciesthat can only survive in traditional,
low-intensityfarmed areas (Hölzinger 1987), our study suggeststhat
the persistence of Wrynecks is possible inareas with intensive
agriculture, as long as theessential resources (access to
ground-dwelling antsand breeding cavities) are present.
We are very grateful to Juliet Vickery, Myles Menz, PaulDonald
and two reviewers for comments on an earlierversion of this
manuscript, as well as to Laura Dafond,Stephanie Geiser, Myles
Menz, Patricia Portner, NatalinaSignorell, Paul Mosimann-Kampe,
Nadja Weisshaupt,Samuel Ehrenbold, Jacques Laesser, Valérie
Coudrainand Fitsum Abadi Gebreselassie for their support
withfieldwork and statistical advice.
REFERENCES
Andrén, H. 1990. Despotic distribution, unequal
reproductive
success, and population regulation in the Jay Garrulus
glandarius. Ecology 71: 1796–1803.
Arlt, D. & Pärt, T. 2007. Nonideal breeding habitat
selection: a
mismatch between preference and fitness. Ecology 88:
792–801.
Benton, T.G., Vickery, J.A. & Wilson, J.D. 2003.
Farmland
biodiversity: is habitat heterogeneity the key? Trends Ecol.
Evol. 18: 182–188.
Bitz, A. & Rohe, W. 1993. Nahrungsökologische
Untersuchun-
gen am Wendehals (Jynx torquilla) in Rheinland-Pfalz. Beih.
Veröff. Naturschutz Landschaftspflege Baden-Württemberg
67: 83–100.
Burnham, K.P. & Anderson, D.R. 1998. Model Selection and
Inference. A Practical Information-Theoretic Approach. New
York: Springer.
Table 3. Model selection results for clutch size (complete
clutches of n = 122 broods from 53 territories), the number
of
fledglings from successful broods (n = 94 from 50
territories)
and the proportion of successful broods (n = 175 broods from
62 territories) in relation to the amount of old pear
orchards
and the number of parcels within the territories, evaluated
with
a generalized liner mixed model with Poisson or binomial
error
distributions. Year is included as a fixed effect in all
models,
and territory as a random effect. Given are the DAIC, AICweights
(wi), number of estimated parameters (K ) and the
deviance.
Model DAIC wi K Deviance
Clutch size
Intercept 0.00 0.404 7 449.00
Old pear orchard 1.94 0.153 8 448.95
Number of parcels 0.46 0.321 8 447.46
Old pear orchard +
number of parcels
2.39 0.122 9 447.40
Number of fledglings
Intercept 0.00 0.401 7 427.70
Old pear orchard 1.56 0.184 8 427.26
Number of parcels 0.66 0.288 8 426.37
Old pear orchard +
number of parcels
2.28 0.128 9 425.99
Proportion of successful broods
Intercept 0.00 0.498 7 248.89
Old pear orchard 1.94 0.189 8 248.83
Number of parcels 1.58 0.226 8 248.46
Old pear orchard +
number of parcels
3.50 0.087 9 248.38
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
740 M. Mermod et al.
-
Coudrain, V. 2009. Food or nesting place? Identifying
factors
limiting wryneck populations. MSc Thesis, University of
Bern.
Danchin, E., Boulinier, T. & Massot, M. 1998.
Conspecific
reproductive success and breeding habitat selection: impli-
cations for the study of colonality. Ecology 79: 2415–2428.
Dauber, J. & Wolters, V. 2004. Edge effects on ant
commu-
nity structure and species richness in an agricultural land-
scape. Biodivers. Conserv. 13: 901–915.
Della Santa, E. 1994. Guide pour l‘identification des
princi-
pales espéces de fourmis de Suisse Centre cartographie
de la faune, Neuchâtel.
Doligez, B., Danchin, E., Clobert, J. & Gustafsson, L.
1999.
The use of conspecific reproductive success for breeding
habitat selection in a non-colonial, hole-nesting species,
the
collared flycatcher. J. Anim. Ecol. 68: 1193–1206.
Freitag, A. 1996. Le régime alimentaire du Torcol
fourmilier
(Jynx torquilla) en Valais (Suisse). Nos Oiseaux 43:
497–512.
Freitag, A. 1998. Analyse de la Disponibilité
Spatio-Temporelle
des Fourmis et des Stratégies de Fourragement du Torcol
Fourmilier (Jynx torquilla L.). PhD Thesis, Université de
Lausanne.
Fretwell, D.S. & Lucas, H.L. Jr 1969. On territorial
behavior
and other factors influencing habitat distribution in birds.
I.
Theoretical development. Acta Biotheor. 19: 16–36.
Geiser, S., Arlettaz, R. & Schaub, M. 2008. Impact of
weather
variation on feeding behaviour, nestling growth and brood
survival in Wrynecks Jynx torquilla. J. Ornithol. 149: 597–
606.
George, T.L. & Zack, S. 2001. Spatial and temporal
consider-
ations in restoring habitat for wildlife. Restor. Ecol. 9:
272–
279.
Glutz von Blotzheim, U.N. & Bauer, K.M. 1980. Handbuch
der Vögel Mitteleuropas. Bd. 9. Columbiformes – Picifor-
mes, 1st edn. Wiesbaden: Akademische Verlagsgesells-
chaft.
Hanski, I. 1998. Metapopulation dynamics. Nature 396: 41–
49.
Holmes, R.T., Marra, P.P. & Sherry, T.W. 1996.
Habitat-spe-
cific demography of breeding black-throated blue warblers
(Dendroica caerulescens): implications for population
dynamics. J. Anim. Ecol. 65: 183–195.
Hölzinger, J. 1987. Die Vögel Baden-Württembergs.
Gefähr-
dung und Schutz. Stuttgart: Ulmer.
Hölzinger, J. 1992. Brutvorkommen des Wendehalses (Jynx
torquilla) an seiner südlichen Arealgrenze in Griechenland.
Beih. Veröff. Naturschutz Landschaftspflege Baden-Württem-
berg 66: 43–46.
Krüger, O. 2002. Analysis of nest occupancy and nest
reproduc-
tion in two sympatric raptors: Common Buzzard Buteo buteo
and Goshawk Accipiter gentilis. Ecography 25: 523–532.
MacKenzie, D.I., Nichols, J.D., Lachman, G.B., Droege, S.,
Royle, J.A. & Langtimm, C.A. 2002. Estimating site occu-
pancy rates when detection probabilities are less than one.
Ecology 83: 2248–2255.
MacKenzie, D.I., Nichols, J.D., Hines, J.E., Knutson, M.G.
&
Franklin, A.B. 2003. Estimating site occupancy, coloniza-
tion, and local extinction when a species is detected imper-
fectly. Ecology 84: 2200–2207.
Muller, K.L., Stamps, J.A., Krishnan, V.V. & Willits,
N.H.
1997. The effects of conspecific attraction and habitat
quality on habitat selection in territorial birds
(Troglodytes
aedon). Am. Nat. 150: 650–661.
Orians, G.H. & Wittenberger, J.F. 1991. Spatial and
temporal
scales in habitat selection. Am. Nat. 137: 29–49.
Pärt, T. 2001. The effects of territory quality on
age-dependent
reproductive performance in the Northern Wheatear,
Oenanthe oenanthe. Anim. Behav. 62: 379–388.
R Development Core Team. 2004. R: A Language and
Environment for Statistical Computing. Vienna, R Founda-
tion for Statistical Computing.
Roth, R. 1976. Spatial heterogeneity and bird species
diversity.
Ecology 57: 773–782.
Royle, J.A. & Dorazio, R.M. 2008. Hierarchical Modeling
and
Inference in Ecology. The Analysis of Data from Popula-
tions, Metapopulations and Communities. New York:
Academic Press.
Sanderson, F.J., Donald, P.F., Pain, D.J., Burfield, I.J.
&
van Bommel, F.P.J. 2006. Long-term population declines
in Afro-Palearctic migrant birds. Biol. Conserv. 131: 93–
105.
Schaub, M. & von Hirschheydt, H. 2009. Effect of current
reproduction on apparent survival, breeding dispersal, and
future reproduction in Barn Swallows assessed by multistate
capture–recapture models. J. Anim. Ecol. 78: 625–635.
Seifert, B. 1996. Ameisen Beobachten, Bestimmen. Augsburg:
Natur Buch Verlag.
Sergio, F. & Newton, I. 2003. Occupancy as a measure of
territory quality. J. Anim. Ecol. 72: 857–865.
Sim, I.M.W., Burfield, I.J., Grant, M.C., Pearce-Higgins,
J.W. & Brooke, M.D. 2007. The role of habitat composi-
tion in determining breeding site occupancy in a declining
Ring Ouzel Turdus torquatus population. Ibis 149: 374–
385.
Stamps, J.A. 1988. Conspecific attraction and aggregation in
territorial species. Am. Nat. 131: 329–347.
Tye, A. 1992. Assessment of territory quality and its effects
on
breeding success in a migrant passerine, the Wheatear
(Oenanthe oenanthe). Ibis 134: 273–285.
White, G.C. & Burnham, K.P. 1999. Program MARK: survival
estimation from populations of marked animals. Bird Study
46: 120–139.
Received 14 October 2008;revision accepted 12 July 2009.
SUPPORTING INFORMATION
Additional Supporting Information may be foundin the online
version of this article:
Appendix S1. Description how to obtain habi-tat-specific ant
nest densities corrected for imper-fect detection.
Table S1. Description and summary statisticsof recorded habitat
types for each of the 100territories.
Table S2. Description and summary statistics ofterritory
variables for each of the 100 territories.
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
Territory quality in Wrynecks 741
-
Table S3. Model selection summary for detec-tion probability of
ant nests.
Table S4. Model selection results for univariatestatic occupancy
models of Wryneck territories.
Table S5. Model selection results for univariateextinction
probability models.
Table S6. Model selection results for univariatecolonization
probability models.
Figure S1. Example of a mapped, digitizedterritory.
Please note: Wiley-Blackwell are not responsiblefor the content
or functionality of any supportingmaterials supplied by the
authors. Any queries(other than missing material) should be
directed tothe corresponding author for the article.
ª 2009 The AuthorsJournal compilation ª 2009 British
Ornithologists’ Union
742 M. Mermod et al.